Volume 17 Number 8 1989
Nucleic Acids Research
In vivo gene expression directed by synthetic promoter constructions restricted to the -10 and -35
consensus hexamers of E.coli
Marie-Ange Jacquet*, Ricardo Ehrlich' and Claude Reiss
Institut Jacques Monod, CNRS and Universite Paris VII, Tour 43, 2 Place Jussieu 75251, Paris cddex 05,
France
Received February 2, 1989; Revised and Accepted March 23, 1989
ABSTRACT
Two synthetic DNA sequences, carrying no other known E. coli promoter element than the consensus
hexamers (CH) TTGACA (CH-35) and TATAAT CH(- 10), spaced by 17 bp, were inserted in
pBR329, in a position enabling transcription of the complete Cmr gene. The region upstream of
the Cmr transcription start was carefully cleared of w.t. promoter elements (full deletion of the wild
type (w.t.) Cmr promoter upstream +2 and large portion of an upstream coding sequence). Both
synthetic promoters, which differ only by the sequences of the spacers (non consensus, constrained
in AT or GC) support in vivo high level Cmr gene expression. The GC rich spacer is associated
with transcription start at the usual + 1 position, but with the AT rich spacer, transcription starts
at several places, mainly in CH(- 10). Rearranged promoter sequences derived from the synthetic
ones upon transformation with partly ligated plasmids, yield new insights on the role of the standard
CH pair, the size of the spacer and the sequence downstream of CH(-10).
INTRODUCTION
Sequence comparison among over 250 promoters recognized by E. coli. RNA-polymerase
(RNAP) reveals (1) a baffling diversity, except for the so-called, -10 and -35 consensus
hexamers (CH) (transcription starts at address + 1). However, the sequences of the CH
can deviate appreciably from the standard CH sequences, TTGACA CH(-35) and
TATAAT CH(- 10); in particular, no known natural promoter exhibits the standard CH
set. On the other hand, sequences matching reasonably well the CH sequences and spaced
by 17 + 2 base pairs (bp) are found at many places in the genome where no promoter
function is detected (2).
Outside the CH sequence, conservation is of poor statistical significance, and point
mutations usually have unpredictable effects. It could well be that the classical 'vertical'
averaging of sequences aligned 'horizontally' is not adequate. Synergic signal elements
could be deposited at given places within an individual promoter, modulating for instance
the effect(s) of the CH pair. These elements could differ, or occur at different places,
for different promoters and hence 'vertical' averaging of the promoters would not allow
their identification (see (3) for examples).
Indications as to the presence of signal elements outside the CH are puzzling. For certain
promoters, sequences upstream of -35 (up to -160) (4) or downstream of -10 (as far
as +20) (5) have been shown to have important effects on promoter strength. In other
promoters, deletion of the region upstream of -35 (6), or even the CH(-35) itself (7)
has little effect on transcription efficiency. Clearly, elements contributing to the promoter
signal can reside throughout the region footprinted by RNAP (typically from -50 to + 10
(8); in some cases footprints extending to -100 or +20 have been observed (9)).
(D IRL Press
2933
Nucleic Acids Research
To investigate the promoter signal, synthetic promoter sequences were tested which
conform to the best average sequences available from a variety of promoter compilations.
They were ligated into plasmids, at places where the core of the w.t. promoter had been
deleted, and tested for their ability to promote transcription. The two largest synthetic
promoter sequences were constructed by Dobryrinin et al. (10) and by Rossi et al. (11);
both were introduced in place of the tetR promoter in pSC101, from CH(-35), either
to the transcription start (11) or to the translation start (10), downstream. Transcription
efficiencies ranked from moderate (Dobrynin's construction, tested by Deuschle et al. (5)),
to excellent (1 1). However, both constructions were tested after insertion into the genome
context of the excised w.t. promoter, and thus unidentified, w.t. promoter signal elements
upstream and downstream of the insert could still be present. Furthermore, both inserts
carried, outside the CH pair, 'best' consensus sequences (although the scores for consensus
never exceeded 40% for any base).
To avoid ambiguities connected with uncontrolled, possibly ubiquitous signal elements,
we decided to test synthetic promoter sequences restricted to the standard CH pair spaced
by 17 bp, surrounded by deliberately non-consensus sequences. In particular, both CH
were framed by unique restriction sites, enabling various combinations of upstream,
downstream of spacer or hexamer sequences to be tested. These were inserted into a genome
site devoid of any known promoter activity. We report here the in vivo behaviour of the
first series of such constructs, aimed primarily at the relation of a standard CH pair to
promoter activity, spacer sequence, and length.
MATERIALS AND METHODS
Bacterial strains and plasmid vector
E. coli strain HB 101 was used as host in transformation experiments ; the plasmid vector
was pBR329 (12).
Synthetic promoters
Both direct and complementary strands of the synthetic promoters (Fig. 1) were synthesized
by the phosphotriester method in a SAMI (Biosearch) DNA synthesizer, and phosphorylated
with 32p using polynucleotide kinase. Both strands were anealed at 37°C overnight, in
220 mM sodium phosphate, 400 mM KCl, 1 mM EDTA.
Plasmid construction
Restriction endonucleases and other enzymes used for nucleic acid engineering were from
commercial sources and used according to the supplier's instructions ; standard molecular
cloning methods were employed (13).
The larger fragment (3804 bp) in the BamH I-Hind III digest of pBR329 (Fig.2) was
isolated by electrophoresis in 2 % agarose. This fragment was mixed with the synthetic
promoter duplex. Ligation with T4 DNA ligase was partial only, as enzyme concentration
and reaction time conditions did not allow completion of the ligation reaction. The mixture
of fully and partly ligated plasmids was introduced into HB1I1 cells made competent by
CaC12 treatment. Transformed cells were isolated on LB-agar plates containing 100tg/ml
ampicillin, after the different recombinant colonies containing the insertion were identified
by screening on chloramphenicol/agar plates at different antibiotic concentrations (100 and
400 /Ag/ml) and by hybridization on nitrocellulose filters. Alternatively, minipreparations
of plasmids were tested for the unique restriction sites which had been introduced into
the synthetic promoter sequence (see Fig.1).
The nucleotide sequences of the Nar I-Pvu II restriction fragment (236 bp) which
2934
Nucleic Acids Research
inserted synthetic sequences
DBR OC- 0and 1e
BamHl -35 Kpnl Smal Xhul
-10 Hind 114 Ciii
5'- GTAGATCC TOACA GGTACGGGCCTCW TATMT MGCTTATCOAT-3'
DBROC-2
5'- GTAGAOGATCC TTOACA
GGTAOCTGTfAC4A
DBROC-4
5'- GTAN0TCC TTCA
---------- TTATCGAT-3'
TCTTATC AT-3'
DBR OC-6
M
5'- OTAAOGATCM C TTOACA
T
OTACCCOOOCCTCXA- IATMT MCTtATAT-3
pBR OC-8
(
5'- GTAGAOGATCC TTeAC GTACIOGCCTCGTTAT/T MOCTTATCOAT-3
DBR AT-O
BuiHi -35
BciI
Xb.l
-10 Hind iII Cibl
5'- GTAGAOATCC TT8CA TCTAGATTATATATCA TATMT AACTTATcGT-3'
DBRAT-I |0
1
5'- OTAIAMATCC TTOA TCTAATTATATATCA TATAT A-CTATQT-3
DBR AT-
TCTAOTTA
5- GTAAGGATCC TTOA TCTAGATTAA'TATCA
l
l
TATiT A-TTATCT-3'
DBR AT-2
5'- OTAGABTCC TTOACA TCTAIATTATATIATCA TATTAACCT
AG
COAicT-3'
CTA TTATGTTCATAT
DOBR AT-3
5- ATOGTC
DBR AT-4
5- GTGA&
T ATTATTATATQIA TfTGA ~tCTTATCAT-3
QTMM
t
TCC TTGACA TCTAGAT--ATGATCA TATMT
AACTTATCGt-3'
DBR AT-S
5'- GTAGAGATCC TTGACA TCTAGATTATATGATCA TATMTA-AGCTTATCtAT-3'
CTAAT0TCMG
DBR AT-6
S' - OTAAOTCC - -- ---
pBR AT-8
5-
- -
TAOATTATATATCA TATMT. -AGTTATCOT-3'
CTABATGTCMG
TCCCGGOCTAGlATOATTCA TATMTA MOCTTATCOAT-3'
CTAOTOTC
Fig.1. Sequences of synthetic and rearranged promoters
The consensus hexamers TTGACA and TATAAT (bold faced) and unique restriction sites of the designed sequences
are indicated. Rearranged bases are underlined, dashes link contiguous bases after deletion.
pBRGC-O and pBRAT-O are the sequences chemically synthesized and cloned into the pBR329 plasmid deleted
of the small Hind HI-BamH I fragment.
pBRGC-2,-4,-6,-8 and pBRAT-l,-2,-3,-4,-5,-6,-8,-10 are rearranged sequences.
A insertion sites.
41 Arrows indicate in vivo transcription start sites, accurate to within 1 base (results
from at least 4 independent experiments).
,
contains the cloned promoter, was obtained by the Maxam and Gilbert procedure (14).
In vivo expression of chloramphenicol resistance
The ability of plasmids to confer resistance to increasing concentrations of chloramphenicol
in liquid cultures was examined by exposing aliquots of a rapidly growing culture to
chloramphenicol (up to 1 mg/ml). Cell growth was monitored by OD5^O. Efficiency of
plating was measured on agar containing 5O,ug/ml ampicillin and 100 itg/ml chloramphenicol; incubation was overnight at 37°C (Fig.3).
2935
Nucleic Acids Research
Cmr
P H
B
(pBR329
v5 b
Apr
Tet.r
ORI
Digestion Hind III-BamHl
Cm.s
r
3804 bp
*7Purification
of the larger fragment
346 bp
Tet3
Conversion to blunt ends
by treatment with
the Klenow fragment of DNA polymerase
,ftt
synthesis of both complementary
oligonucleotides of 31 mer
phosphrylation
Hybridization
of 31 mer-duplex
Dephosphorylation
ligation
c
nsertion of
Pr:3synthetic sequences
3839b Tets
r
r
pBRAT-o
pBRGC-O
Fig.2. Plasmid construction
Strategy used to construct the synthetic promoter sequences and their insertion within pBR329 deleted of Hind
I-BamH I small fragment.
The circular map of pBR329 is shown with the localisation and direction of Apr, Tcr, Cmr genes; restriction
sites used for the sequencing: Nar I (N), Pvu 11 (P); for cloning: Hind Ill (H), BamH I (B) ; ORI, origin
of DNA replication.
Mapping transcription start site in vivo.
A synthetic 20-mer DNA primer (sequence +27 to +47, assuming transcription start of
w.t. Cmr gene in pBR329 in the middle of the Hind HI restriction site (12)) was prepared
as described above. The HPLC-purified primer was end-labeled with 632p -ATP using
T4 polynucleotide kinase (Biolabs).
Preparation of mRNA synthesized in vivo in E.coli strains HB101 carrying the plasmids
under investigation was performed as described in (15). 20 jtg of total mRNA were
2936
Nucleic Acids Research
coprecipitated with 1. 107 cpm of the primer. The precipitate obtained after spinning in
an Eppendorf centrifuge (4°C) was resuspended in 20 1l of hybridization buffer (80 %
formamide, 0.4 M NaCl, 1 mM EDTA, 0.04 M Pipes, pH 6.4). Incubation was at 47°C
overnight, after preincubation for 2' at 85°C. Hybrids confined in the aqueous phase
following phenol treatment were precipitated with ethanol. Primer extension was carried
out (42°C, 60 minutes) with 30 ,ml AMV reverse transcriptase (Boehringer Mannheim),
300 units/ml, tris HCl, pH 8.3, KCl 140 mM, MgCl2 10 mM, mercaptoethanol 20 mM;
extended products were migrated in 10 % polyacrylamid gel, 7 M urea together with size
markers, and analyzed by autoradiography (autoradiograms not shown ; as reported in
(16), the size of the markers were shifted 1-2 bp upwards with respect to the extended
products).
RESULTS
Synthetic promoter sequences
Two synthetic promoter sequences were designed (Fig. 1), sharing the same length (3 lbp),
both standard CH (1TGACA at -35 and TATAAT at -10) spaced by 17 bp, and restriction
sites BamH I, just upstream of CH(-35), and Hind HI, just downstream of CH(- 10).
They differ in their spacer sequence: GC-0 has a spacer of 76.5 % GC which carries
unique restriction sites for Kpn I, Sma I and Xho I ; AT-0 has a spacer of 76.5 % AT
which carries unique restriction sites for Xba I and Bcl I. The spacer sequences were chosen
to avoid possible formation of particular structures (cruciform, bends, etc...). In the two
constructions, both CH are tightly framed by unique restriction sites ; these synthetic
promoters are therefore convenient construction sets for building promoters from a variety
of sequence elements.
Promoter insertion into pBR329
The synthetic promoters were cloned into pBR329, replacing the 346 bp long Hind
mI-BamH I fragment. The BamH I site in pBR329 (address 606 in the sequence numbering
of Covarrubias and Bolivar (12)) covers codon 96 in the Tcr gene (Fig.2) ; Hind m
(address 260) cleaves the transcribed sequence of the Cmr gene at position +2.
Consequently, the removed Hind mI-BamH I fragment of 346 bp carries the promoter
and transcription initiation site controlling expression of Cmr on one strand, and the Tcr
gene from CH(-10) of the Tcr promoter to about 1/4 of the Tcr coding sequence (Nterminal side), on the other strand. The larger (3804bp) Hind Il-BamH I circularized
fragment of pBR329 was indeed found to confer Apr/Cms and Tcs to HB1I1 (data not
shown).
Both synthetic promoters were ligated into the Hind II/BamH I gap of pBR329 (ligation
reaction not completed) and the resulting plasmids used to transform E. coli strain HB1O1.
Tcs/Apr transformants (carrying the plasmids) were identified by colony hybridization and
screened for the Cmr phenotype. The presence of the synthetic promoters in the
transformants was determined by mapping the unique restriction sites they carry.
Two transformants were recovered, carrying the designed promoter sequences. pBR GC-1
has exactly the planned GC-0 sequence ; pBR AT-10 has the planned AT-0 insert, except
for the deletion of an A in the Hind III site downstream of CH(-10); the larger Hind
HI-BamH I fragment of pBR329 is found unmodified in both plasmids.
Promoter sequence rearrangements.
As stated in Materials and Methods, the ligation reactions with both designed sequences
AT-0 and GC-0, were not brought to completion and hence transformation was attempted
2937
Nucleic Acids Research
i
0
L-
9
9
d
Fig.3. Efficiency of plating
Growth on LB-agar plates of HB1O1 cells carrying (1) no plasmid, ; (2) pBR329 deleted of small fragment Bam
HI-Hind III and ligated; (3) pBR329, (4) pBRGC-l, (5) pBRGC-2, (6) pBRGC-4, (7) pBRGC-6, (8) pBRGC-8,
(9) pBRAT-l, (10) pBRAT-2, (11) pBRAT-3, (12) pBRAT-4, (13) pBRAT-5, (14) pBRAT-6, (15) pBRAT-8.
A-Control plates with 50 Isg/ml ampicilin
B-Plates containing 100 jsg/ml chloramphenicol respectively. a) to d) correspond to dilutions of cell stock (from
stationary culture, 100, 10-2, 10-4 and 10-6 times, respectively; 10 ul of cell suspension were deposited in
each case.
with a mixture of linear and circular plasmids in each case. Transformation with linear
DNA is known to be inefficient, and to generate high mutation rates in the vicinity of
the DNA extremities, in contrast to circular DNA (17). Indeed, for both pBR AT and
pBR GC series, the transformation yield, as deduced from the Amp' phenotype, was low
(1O-4). Among the transformants, a large spectrum of Cm resistance was observed. For
both series, a set of colonies was isolated, each displaying a different level of Cm resistance
2938
Nucleic Acids Research
100BC
4)
20[
100
300 5s0
800 1000 100 300 500
chloramphenicol (pg/mi)
800 1ooo
Fig.4. Chloramphenicol resistance in liquid medium for HBO1I cells containing different plasmids.
Top: A-Controls: with HB101 cells no plasmid ;HB101 carrying pBR329 deleted of Hind-Bam HI (346 bp)
A and pBR329 *.
Bottom : B-Cells containing the plasmids of pBRGC series : pBRGC-1:V; pBRGC-6: L]; pBRGC-8:E;
pBRGC-2: *; and pBRGC-4: V.
C-Cells containing the plasmids of pBRAT series: pBRAT-10: A; pBRAT-5: *; pBRAT-1: 0; pBRAT-2: *;
pBRAT-4:V; pBRAT-3:0; pBRAT-6:* and pBRAT-8:V.
Ordinate: % of cell mass with respect to untreated cells, mesured by optical density at 550 nm after 2 hours
growth in LB medium at 370 of an initial 0.05 OD550 culture.
and cartying at least one of the unique restriction sites introduced into the synthetic promoter;
plasmids were sequenced between Pvu II and Nar I.
A great variety of sequences rearranged in the vicinity of the ligation sites was obtained.
Those we have sequenced (Fig. 1) are likely to be a small sample of the rearranged sequences
actually produced. For the pBR GC series, among the variants analyzed, only pBR GC-8
keeps both CH, but its spacer is increased to 18 bp and a C to T transversion had occurred
in the Kpn I/Sma I site. The 3 other plasmids isolated, pBR GC-2,-4 and -6 carry CH(-35),
but only the latter has a remnant (5/6) of CH(- 10). None of the pBR GC series had
rearranged in the promoter area upstream of Kpn I or downstream of Hind III.
For the pBR AT series, 8 promoter sequences were characterized ; 7 members of the
series carry CH(- 10), 5 have in addition CH(-35), one has none (pBR AT-3) ; three
2939
Nucleic Acids Research
Table 1. Chloramphenicol resistance of selection plasmids
Cm concentrations at which 50% of the HB 101 cells carrying plasmids as indicated survive in LB liquid medium.
Data are average over 5 independent experiments at least.
Plasmid
pBR
pBR
pBR
pBR
pBR
GC
GC
GC
GC
GC
-1
-6
-8
-2
-4
pBR 329
Cm concentration
50% HBIOI survival
Plasmid
650 Ag/ml
530 pg/mI
490 tzg/ml
70 zg/ml
34 jg/ml
pBR AT -10
pBR AT -5
pBR AT -1
pBR AT -2
pBR AT -4
pBR AT -3
pBR AT -8
pBR AT -6
350
jig/ml
HBIOI cells,
no
pBR 329, deleted
of small BamHIHind III fragment
and ligated
Cm concentration
50% HBIO survival
650
600
490
360
320
140
50
26
jg/ml
yg/ml
Ag/ml
ikg/ml
uLg/ml
Lg/ml
jg/ml
/g/ml
10 yg/ml
plasmid
20 ,sg/ml
of the five promoters which have both CH also maintained the synthetic spacer sequence
(pBR AT-10, 2, 5); one has 4 bases deleted from the spacer (pBR AT-4) and one has
inserted a repeat of 9 bases (pBR AT-1). pBR AT-4 is the only promoter not having a
rearranged synthetic sequence downstream of CH(- 10) ; all others have lost the Hind
Ill site flanking CH(- 10). Four of them have inserted next to CH(- 10) a repeat (9 to
19 bp) taken from the spacer sequence and part of CH(- 10). Although the main focus
in the present contribution is on the designed sequences, pBR GC-l and pBR AT-10, we
include some relevant results obtained with rearranged sequences.
In vivo activity mediated by the selected promoters.
Fig. 3 and 4 show the growth of HBO11 cells transformed by the two selected series of
plasmids in solid agar and liquid medium containing increasing chloramphenicol
concentrations. 50 % inhibition is reached at chloramphenicol concentrations indicated
in Table 1.
In both series, promoters with synthetic CH(- 10) and CH(-35) display in vivo Cm
resistances 1-2 times that of w.t. pBR 329. The levels of resistance conferred by the
corresponding plasmids are not sensitive to spacer sequence constraints (76 % GC or 76
% AT) and poorly to spacer length (14 to 26 bp). A 5-10 fold reduction of Cm resistance
is observed with promoter sequences having lost one CH, either CH(- 10), apparently
specific of the pBR GC series, or CH(-35), which occurs in the pBR AT series.
The behaviour of pBR AT-3, which has no CH, is surprising, as it still confers appreciable
Cm resistance to its host (40 % that of w.t. pBR 329).
In vivo transcription start sites for selected promoters.
Mapping transcription start sites was attempted by the primer extension assay using AMV
reverse transcriptase. The synthetic primer was complementary to the 20 base sequence,
positions +27 to +47 of the Cmr mRNA in pBR329 (assuming transcription start (address
+ 1) in the middle of the Hind III restriction site (12)) ; the initiation codon of the Cmr
gene starts at +48.
2940
Nucleic Acids Research
Transcription start sites could be mapped by this method only for the class of promoter
constructions conferring moderate to high chloramphenicol resistance to the host bacteria,
i.e. maps for pBR GC-2 and -4, and pBR AT-6 and -8, could not be obtained. The results
are displayed on fig. 1. pBR GC-1 starts transcription in vivo 6 bp downstream of the
CH(- 10), so is similar to w.t. pBR329, which initiates transcription at two sites located
5 (major site) and 7 bp downstream of the CH(- 10) hexamer. pBR GC-6 has its main
start site also 6 bp downstream of the CH(- 10). pBR GC-8 is unconventional, as its
transcription starts mainly within CH(- 10). Still more unexpected, pBR AT-1 and -10
initiate transcription at three sites, one 4bp downstream of the CH(- 10), but the two others,
including the major middle one, are in CH(-10). The other rearranged sequences of the
pBR AT series display mainly single transcription start sites, pBR AT-2 in CH(- 10),
pBR AT-3 7bp, pBR AT-4 12bp, and pBR AT-5 7bp, downstream of CH(- 10). For each
plasmid, the primer extension assay was repeated at least 4 times and gave consistent results.
No major start site was found downstream of the Cla I site.
All Cmr mRNA genes transcribed in vivo by the class of promoter constructions
considered here carry most-if not all-untranslated sequences of w.t. Cmr mRNA,
involved in ribosome attachment and assembly. It is therefore very likely that mRNA species
produced by this class of promoter constructions are translation templates of equivalent
efficiency. This is further supported by our preliminary results of mRNA quantitation in
vivo which show for this class of promoter constructions an almost constant ratio of the
amount of mRNA to antibiotic resistance as given in table I (M.A.Jacquet, unpublished).
,
DISCUSSION AND CONCLUSIONS
Design and test of 'modular' promoters.
A promoter can be envisioned as a series of instructions needed for transcription: tight
binding of RNAP, closed, then open complex formation, start address, initiation frequency,
regulatory instructions, etc... It is likely that each instruction is best expressed by a specific
sequence, but slight alterations of the latter would not supress the instructions, only reduce
its effect to some extent.
Control and regulation of expression of a particular gene sets specific demands on its
promoter, which could be met by selecting a given set of promoter instructions. The set
could be hierarchized by suitable alterations of the sequences encoding the instructions.
We believe that this, and the squeezing due to the finite promoter size, are the main reasons
for the observed promoter sequence diversity.
Adopting this view, one could try to take apart the set of instructions of a natural promoter
and study individually each instruction for its role in that promoter. We took a
complementary approach, by introducing, in a sequence environment as neutral as possible,
one or a few putative promoter instructions and check for their possible ability to express
a gene in vivo, then to study the molecular mechanism of their activity, if any. To this
end, we designed sequences carrying both standard CHs, framed with unfrequent restriction
sites enabling easy handling of these or of other known or anticipated promoter elements.
Once ligated upstream of a selected test gene, the construction can be tested for its ability
to promote gene expression. To be conclusive, testing of synthetic promoters or promoter
elements should be performed in a DNA context devoid of uncontrolled promoter activity
; this requirement is not necessarily met-and difficult to control-if the synthetic sequence
is inserted in, or in the context of, a w.t. promoter, as was the case for most studies of
synthetic promoter or promoter elements published so far. Since it is at present difficult
2941
Nucleic Acids Research
to recognize promoter elements outside of the -10 to -35 promoter core, the safest is to
remove as much sequence up and downstream of the w.t. promoter as compatible with
in vivo expression of the gene, prior to introducing the construction to be tested.
The primary goal of the present study was twofold: is a pair of standard CH spaced
by 17bp sufficient information to promote transcription ? What is the role of the sequence
of the 17bp spacer ? Unambiguous answers are provided by the planned constructions,
pBR GC-1 and pBR AT-10.
A pair of CH, spaced by 17 bp, is sufficient for strong gene expression in vivo.
When ligated in front of the Cmr gene, at its transcription start site, both pBR AT-10 and
pBR GC-1 support indeed very efficient gene expression in vivo, almost twice that of the
w.t. promoter they replace. This confirms and quantitates previous findings, for instance
those obtained by inserting, in place of the core of the w.t. promoter of Tetr, a sequence
carrying the standard CH pair, spaced by a 17 bp 'consensus' sequence (11).
Our results go however one step further, as our constructions deliberately avoid
consensus' in the spacer sequence, and operate in a sequence context from which w.t.
promoter elements have been removed to the best of our knowledge, at least upstream
of + 1. We conclude that a pair of CH sequences, spaced by 17 bp, is sufficient by itself
for strong gene expression in vivo. This statement does not exclude that putative late
promoter elements (5), which could be deposited in the transcribed 5' end of the Cmr
gene and hence would be present in our constructions, could contribute to the expression
(see below).
The sequence of the 17 bp spacer and sequence downstream of CH(- 10) involved in setting
transcription start.
Our laboratory had been previously active in analyzing composition constraints in natural
DNA sequences (AT/GC, or purine/pyrimidine (Pu/Py) constraints). The constraints specify
well delimited domains displaying dedicated physical properties (18). When applied to
E. coli promoter sequences, a consensus domain map is found, in which the two domains
harboring the CH pair (obviously rich in AT) were separated by a spacer domain higher
in GC (19). We therefore decided, as a first application of our modular promoter concept,
to synthesize promoter sequences carrying both standard CH, separated by a spacer of
17 bp of either high GC or high AT contents, to see the effect of these composition
constraints on activity. Particular features such as runs of Pu (in particular A), Pu/Py
alternance (specially GC) or palindromic sequences were avoided because of associated
unusual conformations.
Both pBR AT-10 and pBR GC-1 support the same high level of gene expression in vivo.
Thus, for a spacer of 17 bp, AT or GC constraints do not affect expression efficiency
in vivo, a conclusion holding at least for our spacer sequences. However, this conclusion
may be valid only for promoters bearing the standard CH pair, as it was shown by Auble
et al. (20) that the sequence of a 17 bp spacer, linking non-standard CH pairs, can modulate
gene expression in vivo.
A striking difference between pBR GC-l and pBR AT-10 is the location and homogeneity
of their Cmr transcription start sites. We observe that close to 100 % of the Cmr
transcripts in pBR GC-1 start at the conventional place, 6 bp downstream of CH(-10).
In pBR AT-10, Cmr transcription start is scattered over a 9 bp sequence, and occurs at
3 main sites; about 2/3 of the transcripts start in CH(- 10) (result not shown, our unpublished
observations). This surprising location has been consistently found in 10 separate primer
extension experiments carried out with Cmr transcripts from pBR AT-10. Systematic
2942
Nucleic Acids Research
,
100
X
x.'
90
80
,X
70
1)~~~~~~~~~~~~~~~~~~~~~~~%
°
60
=3
50
I
C
0
0'
40
ii
30
20
10
13
14
15
17
18
16
size of spacer (bp)
19
20
Fig.5 Gene expression
Level of gene expression conferred by a synthetic promoter construction bearing a standard CH pair, as a function
of spacer size (relative to 17 bp spacer taken as 100 %).
* present work
x constructions of Aoyama et al. (24) tested in (26)
* taken from Brosius et al. (25)
artifacts in these experiments are unlikely, as in other cases (PBR329 for instance), they
gave the expected start sites. As pBR GC-1 and pBR AT-10 differ only in their spacer
sequence (and by a 1 bp deletion downstream of CH(-10) in the latter), the results draw
attention on contribution from the spacer sequence in setting transcription start.
Additional information on start site determinants is provided by the rearranged sequences,
pBR AT-2 and -5, which differ from pBR AT-10 only by the sequences linking CH(-10)
to the ClaI site. Their transcription starts differ from those found in pBR AT-10, as pBR
AT-2 starts in CH(- 10), but at a single site mainly, and pBR AT-5 at the conventional
place, 7 bp downstream of CH(- 10). We are left with the conclusion that for promoters
carrying the standard CH pair, transcription start is not under exclusive control of the
pair, in particular of CH(-10); the sequence of the 17 bp spacer, and/or that downstream
of CH(- 10) are also relevant elements. Contribution of sequences downstream of CH(- 10)
have been stressed earlier (5).
Recent findings reported by Horwitz and Loeb (21), who observe that modified promoter
(with non-standard CH pairs) can have transcription start shifted towards CH(-10),
ressemble our findings with pBR AT-10, -2 (and also pBR AT-1 and pBR GC-8). They
attribute the effect to cryptic CH elements. Surrogate (-10) elements can be found in the
AT rich spacer of pBR AT-10 like TAGATT, TTATAT or TATGAT, but the associated
2943
Nucleic Acids Research
(-35) sequence would match at best the GA pair of CH(-35), which is present also-but
not active-in pBR GC-1. These surrogate elements are shared by pBR AT-5 but are not
active there. It is furthermore difficult to concile the dominance of sequences with poor
fit to consensus over a pair of standard CH, and the generally accepted (but lately challenged
(22)) rule 'consensus is strength' (see the discussion in (21)).
Effects of size of spacers linking a pair of standard CH.
Transformation with partially ligated DNA proves to be a valuable trick for generating,
in the vicinity of the ligation site, a series of randomly located mutations of variable extension
in addition to the designed sequence. The rearranged sequences can be screened for a
selected element, like for instance the size of the spacer linking a pair of standard CH.
This is the case for rearranged plasmids pBR AT- 1, -4 and pBR GC-8, -6 (the latter with
a transversion in CH(-10)) which have spacer size ranging from 14 to 26 bp. The plot
of in vivo gene expression vs spacer size (Fig.5) confirm that 17 bp is the optimum size,
but evidences in addition three novel features: (i) efficiency drops steadily on both sides
of the 17 bp optimum, more abruptly upon increasing than reducing the spacer size. (ii)
for spacers of given size, the precise sequences seems not to affect relative effiency, which
is consistent with our conclusion in the previous section (iii) most importantly, the drop
is rather milde, since mutants with 14 to 26 bp spacers display a twofold variation of gene
expression only ; this is in strong contrast with the effect observed in natural promoters
(with non-standard CH), where departure from the natural 17 bp spacer by as little as
i 1 bp may reduce gene expression by as much as an order of magnitude (20,23,24).
As shown earlier, the spacer size is a negative promoter modulator our results suggest
that its strength can be tuned by the fit of the consensus hexamers to the standard the
better the fit, the smaller the modulation effect.
Sequences of the CH type not necessary for efficient gene expression
Among the rearranged promoter sequences selected, fig. 1, those having deleted one CH,
i.e. pBR GC-2, pBR GC-4, pBR AT-6 and pBR AT-8, still confer Cm resistance, but
their efficiencies have dropped one order of magnitude below that conferred by promoters
bearing the CH pair. This result is consistent with earlier studies (27,28), although results
reported recently (21) show that removal of one CH from a w.t. promoter does not
systematically result in a drop of promoter strength, possibly through rescue from some
cryptic CH element. A first example of a sequence displaying good promoter activity (40%
w.t.), but missing discernable CH elements both 10 and 35 bp upstream of transcription
start, is provided by the rearranged sequence pBR AT-3. Transcription of the Cmr gene
starts in pBR AT-3 in Clal, at the boundary with HindIll. The corresponding -10 and
-35 hexamers read TGTCAA and GTAGAG, respectively. The latter, completely out
of consensus (TTGACA) is borrowed from the Tcr coding sequence just downstream of
BamH I. The former is of unknown origin; it shares 3 bases with CH(-10) (TATAAT),
but misses both most conserved ones (first A and last T). This suggests that the promoter
signal(s) carried in place of the CH pair can be mimicked efficiently by sequences, or
sequence combinations, altered from consensus beyond recognition. Cm resistance conferred
by pBR AT-3 is stronger than that observed for inserts carrying one standard CH only,
but less than that of inserts keeping both. This is strong indication that efficient
supplementing of one CH depends stringently on the sequence occuring at the place of
the other: whether consensus or not, these sequences seem to come in efficient constitutive
promoters as matched pairs, indicative of the synergic action of these signals. The identity
of the cryptic (synonymous) signal(s) involved is at present unknown.
2944
Nucleic Acids Research
Work is is progress to further characterize the promoter activity of BR GC-1 and pBR
AT-10 in vivo, and of selected rearranged sequences. These extreme simplified but active
promoters sequences could help disclosing the basic feature of promoter recognition,
complex formation and activation (M.A.Jacquet, work in progress).
ACKNOWLEDGEMENTS
We wish to thank Prof. R.M.Hochstrasser for comments, Dr Peter Brooks for help with
the manuscript, Prof. T.Aoyama for gift of the constructions described in (24)(see fig.5),
C. Michon-Dubucs for oligonucleotide synthesis, L. Corme for typing the manuscript and
R. Schwartzmann for photographical work.
*To whom correspondence should be addressed
'On leave of absence; present address: Instituto de Investigaciones Biologicas Clemente Estable, Avda.
Italia 3318, Montevideo and Facultad de Humanidades y Ciencas, Tristan, Narvaja 1674, Montevideo,
Uruguay
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
Harley,C.B. and Reynolds,R.P. (1987), Nucl.Acid Res., 15, 2343-51.
Stormo, G.D., Schneider, T.D., Gold, L. and Ehrenfeucht, A. (1982), Nucl. Acid Res., 10, 2997-3011.
Lanzer,M. and Bujard,H. (1988), Proc.Natl.Acad.Sci. USA, 85, 8973-77.
Bossi,L. and Smith,Y.M. (1984), Cell, 39, 643-52
Deuschle, U., Kammerer, W., Gentz, R. and Bujard, H. (1986), Embo J., 5, 2987-94.
Yu,X.M. and Reznikoff,W.S. (1986), J.Mol.Biol., 188, 545-53.
Okamoto,T., Sugimoto,K., Sugisaki,H. and Takanami,M. (1977), Nucl. Acid Res., 4, 2213-22.
Duval-Valentin,G. and Ehrlich,R. (1987), Nucl.Acid Res., 15, 575-94.
Duval-Valentin,G. and Ehrlich,R. (1988), Nucl.Acid Res., 16, 2031-44.
Dobrynin,V.N., Korobko,V.G., Severtson,A.I., Bystrov,N.S., Chuvpilo,S.A. and Kolosov,M.N. (1986),
Nucl.Acid Res. Symp.Ser., 7, 365-75.
Rossi,J.J., Soberon,X., Marumoto,Y., Mc Mahon,J. and Itekura,K. (1983), Proc.Natl.Acad.Sci. USA,
80, 3203-07.
Covarrubias,L. and Bolivar,F. (1982), Gene, 17, 79-90.
Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982), Molecular Cloning: A Laboratory Manual, Cold Spring
Harbor Lab.
Maxam,A.M. and Gilbert,W. (1980), Methods in Enzymol., 65, 499-560.
Jacquet,M.A. and Ehrlich,R. (1985), Biochimie, 67, 987-97.
Ernoult-Lange,M. and May,E. (1983), J.Virol., 46, 759.
Conley,E.C., Saunders,V.A. and Saunders,J.R. (1986), Nucl. Acid Res., 14, 8905-17.
Marcaud,H., Gabarro-Arpa,J., Ehrlich,R. and Reiss,C. (1986), Nucl.Acid Res., 14, 551-58.
Ehrlich,R., Marin,M., Larousse,A., Gabarro-Arpa,J., Schmitt,B. and Reiss,C. (1984), Folia Biologica,
105-18.
Auble,D.T., Allen,T.L. and DeHaseth,P.L. (1986), J.Biol.Chem., 261, 11202-06.
Horwitz,M.S.Z. and Loeb,L.A. (1988), J.Biol.Chem., 263, 14724-14731.
Knaus,R. and Bujard,H. (1988), EMBO J., 7, 2919-2923.
Stefano,J.E.and Gralla,J.D. (1982), Proc.Natl.Acad.Sci.USA, 79, 1069-72.
Aoyama,T., Takanami,M., Ohtsuka,E.,Tanaiyama,Y.,Marumoto,R., Sato,H. and Ikemara,M. (1983),
Nucl.Acid Res., 11, 5855-64.
Brosius,J. Erfle,M. and Storella,J. (1985), J.Biol.Chem., 260, 3539-41.
Jacquet,M.A., (1987), Doctoral thesis, University ParisVII.
Ponnambalam,S., Webster,C., Bingham,A. and Busby,S. (1986), J.Biol.Chem., 261, 16043-48.
Keilty,S. and Rosensberg,M. (1987), J.Biol.Chem., 262, 6389-95.
2945